메뉴
Apply five concrete heuristics to detect tautological tests, mock-rubber-stamps, and tests written after production code. Use after writing a test, before running it, and any time a test passes too easily.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when starting any TDD cycle, before writing any test file, editing any production file, running tests, or advancing a phase — required before any of those actions.
How to drive the vitest-agent MCP tools correctly — the non-obvious operational facts (run_tests scoping, why subset runs "fail" coverage, the consoleLeaks signal, the VITEST_AGENT_CONSOLE escape hatch, auto-recovered attribution) that otherwise cost trial-and-error.
Guide @vitest-agent/plugin configuration including thresholds, targets, output format, and plugin options. Use when setting up or modifying the plugin configuration.
Required before any production-code edit during the red phase. Forces externalization of "why I think this fix will work" with cited test_error_id and stack_frame_id evidence.
Use when starting a dogfood session for the vitest-agent plugin in this repo, when continuing one from an existing handoff at docs/superpowers/dogfood/<chain>/, or after the TDD orchestrator subagent completes a dogfood task. Repo-internal — does not apply to projects that consume vitest-agent as a dependency.
Use when reaching any TDD phase transition — spike close, red→green, green→refactor, or discard — required before exiting the phase so each cycle lands as a discrete, greppable commit in git history.
| name | verify-test-quality |
| description | Apply five concrete heuristics to detect tautological tests, mock-rubber-stamps, and tests written after production code. Use after writing a test, before running it, and any time a test passes too easily. |
Apply these five heuristics to the test you just wrote (or are reviewing). Any that triggers warrants a closer look.
Test imports SUT but only asserts on mock return values. If the test mocks the SUT's collaborators and then asserts that the SUT returns what the mocks returned, the test isn't testing the SUT — it's testing the mock framework. Restructure so assertions land on values the SUT transforms, not values it passes through.
Test body has no expect calls on SUT-returned data.
Tests that only assert on mocks (expect(mockFn).toHaveBeenCalledWith(...)) without ever inspecting what the SUT returned are testing the call shape, not the behavior. Add an assertion on the SUT's return value.
Test name matches /should call/ or /should return what mock returns/.
These are red flags. Rename to describe the observable outcome. (See: derive-test-name-from-behavior.)
Test was created AFTER the production code it claims to test.
Use file_edits ordering to check. If the test file was edited after the production file in the same TDD cycle, the test was written to match the implementation — the test cannot fail in a way that catches the implementation being wrong. The killer signal.
Coverage delta of new test is zero. A test that doesn't increase coverage either duplicates an existing test or asserts something the type system already enforces. Re-evaluate.
Code review and coverage-gap workflows use this. The PostToolUse post-tool-use/test-quality.sh hook (orchestrator-scoped) is a narrow automated check today: it scans test-file edits for escape-hatch tokens (it.skip/it.todo/it.fails/it.concurrent, the test.* and describe.* equivalents, and .skipIf/.todoIf) and records tdd_artifacts(kind='test_weakened') when any of them appear. Heuristics #1, #2, and #4 — mock-rubber-stamping, no-SUT-assertion, and test-after-code — are NOT detected by the hook; they remain agent/reviewer responsibilities and are why this skill is consulted manually after writing a test. Heuristic #3 (suspicious test names) is a manual review item too. Heuristic #5 (zero-coverage-delta) requires a coverage run to evaluate.
